fix: truncate job history dir path length

Signed-off-by: Joel Wong <[email protected]>

src/deadline/client/job_bundle/__init__.py

@@ -34,10 +34,6 @@ def create_job_history_bundle_dir(submitter_name: str, job_name: str) -> str:
         char for char in submitter_name if char.isalnum() or char in " -_"
     )
 
-    # Clean the job_name's characters and truncate for the filename
-    job_name_cleaned = "".join(char for char in job_name if char.isalnum() or char in " -_")
-    job_name_cleaned = job_name_cleaned[:128]
-
     timestamp = datetime.datetime.now()
     month_tag = timestamp.strftime("%Y-%m")
     date_tag = timestamp.strftime("%Y-%m-%d")
@@ -53,8 +49,17 @@ def create_job_history_bundle_dir(submitter_name: str, job_name: str) -> str:
         latest_dir = existing_dirs[-1]
         number = int(os.path.basename(latest_dir)[len(date_tag) + 1 :].split("-", 1)[0]) + 1
 
-    result = os.path.join(
-        month_dir, f"{date_tag}-{number:02}-{submitter_name_cleaned}-{job_name_cleaned}"
-    )
+    job_dir_prefix = f"{date_tag}-{number:02}-{submitter_name_cleaned}-"
+
+    # max path length - manifest file name
+    # 256 - len("\manifests\d2b2c3102af5a862db950a2e30255429_input")
+    # = 207
+    max_job_name_prefix = 207 - len(os.path.abspath(os.path.join(month_dir, job_dir_prefix)))
+
+    # Clean the job_name's characters and truncate for the filename
+    job_name_cleaned = "".join(char for char in job_name if char.isalnum() or char in " -_")
+    job_name_cleaned = job_name_cleaned[: min(128, max_job_name_prefix)]
+
+    result = os.path.join(month_dir, f"{job_dir_prefix}{job_name_cleaned}")
     os.makedirs(result)
     return result

test/unit/deadline_client/job_bundle/test_job_history_folders.py

@@ -5,6 +5,7 @@
 """
 
 import os
+import sys
 import tempfile
 import pytest
 
@@ -55,6 +56,63 @@ def test_create_job_bundle_dir(fresh_deadline_config):
         assert sorted(os.listdir(os.path.join(tmpdir, "2023-04"))) == EXPECTED_DIRS[3:]
 
 
+def test_job_name_truncation_from_long_job_history_dir(fresh_deadline_config):
+    # Use a temporary directory for the job history
+    with tempfile.TemporaryDirectory() as tmpdir, freeze_time("2024-12-27"):
+        long_tmpdir = os.path.join(
+            tmpdir,
+            "dir1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567",
+        )
+        config.set_setting("settings.job_history_dir", long_tmpdir)
+        long_tmpdir_path_length = len(os.path.abspath(long_tmpdir))
+        assert 128 < long_tmpdir_path_length < 196  # smoke test for path length
+        assert os.path.isdir(long_tmpdir)  # smoke test that path was created
+
+        output_path = job_bundle.create_job_history_bundle_dir(
+            "MySubmitter",
+            "job123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890",
+        )
+
+        assert len(os.path.abspath(output_path)) == 207
+        assert str(os.path.abspath(output_path)).startswith(
+            os.path.abspath(os.path.join(long_tmpdir, "2024-12-27-MySubmitter-job1234567890"))
+        )
+        assert (
+            len(
+                os.path.abspath(
+                    os.path.join(output_path, "manifests", "d2b2c3102af5a862db950a2e30255429_input")
+                )
+            )
+            == 256
+        )
+        assert os.path.isdir(output_path)
+
+
+def test_job_name_truncation_from_long_job_name(fresh_deadline_config):
+    # Use a temporary directory for the job history
+    with tempfile.TemporaryDirectory() as tmpdir, freeze_time("2024-08-26"):
+        config.set_setting("settings.job_history_dir", tmpdir)
+
+        job_name_exceeding_128_chars = "test12345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345"
+        output_path = job_bundle.create_job_history_bundle_dir(
+            "SubmitterFour",
+            job_name_exceeding_128_chars,
+        )
+        if sys.platform == "darwin":
+            assert (
+                str(output_path)
+                == str(
+                    os.path.join(tmpdir, f"2024-12-27-SubmitterFour-{job_name_exceeding_128_chars}")
+                )[:207]
+            )
+        else:
+            assert output_path == os.path.join(
+                tmpdir,
+                "2024-12-27-SubmitterFour-test1234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234567890123456789012345678901234",
+            )
+        assert os.path.isdir(output_path)
+
+
 @pytest.mark.parametrize(
     "submitter_name, job_name, freeze_date, expected_output_path",
     [